I recently had 20 drives across three machines. I was using a combination of raid5, iscsi, mhddfs, and samba. Machine1 mounts the iscsi devices from the other two machines, and then mhddfs combines them into one virtual filesystem. Samba is then used to share files out with laptops.

What I found is that network card drives in 3.2 kernels are currently in a horrible state. They crash left and right under real load. This is after trying different brands, tweaks, version of drivers, etc. In addition it seems iscsi client in Fedora 16 is also not in a great state. Independent of network issues, I would still get failures. The machine running CentOS 6.2 used to be Fedora 16, but was converted to make things more stable.

My latest plan is to do basically the same thing I was doing before, but on a smaller scale. I am going to retire the 1tb drives in Machine3, and replace them with the 2tb drives from Machine1. I am also going to convert Machine3 to CentOS 6.2 for stablility. Then Machine2 will mount Machine3's iscsi device, and use mhddfs and samba. This reduces the number of machines involved from three to two, and takes Fedora 16 out of the mix. It will also reduce the number of drives involved in mass storage from 20 to 14.

I plan to add two 2tb drives to Machine1, for storage, but it end up being only a desktop.

Machine1
Fedora 16
6x2tb raid5

Machine2
CentOS 6.2
5x1.5tb raid5
3tx2tb raid5

Machine3
Fedora 16
6x1tb raid5

I hear you, it does feel like a downgrade. On the other hand, grub1 is not working for me. I upgraded to Fedora 16 last night. At first GRUB2 gave me simply "GRUB", and GRUB1 gave me "Error 16". I tried multiple tricks to get GRUB1 working, and was unsuccessful. What I finally ended up having to do was use GRUB and make the empty space at the beginning of the drive 2047 blocks instead of the previous 62. To do this I had to backup the contents of /boot, repartition, redo raid1, format it, and copy the data back.

I also recently ran into the Error 16 error with GRUb1 on Fedora 15 on my mail/web server. To workaround it I ended up installing GRUB2 from Fedora 16.

I am doing very much the same thing. I have six 1tb hard drives in my main desktop, and five 1.5tb in a iSCSI server. I then combine them with mhddfs. It is slow, but I only use it for big files that I am not going to be rewriting. I use linux software raid5 for the big filesystems, and linux software raid10 for my /home.

I am excited to see 4-5tb drives coming down the pipe. With just four 5tb drives I could replace all my hard drives, and remove the need for the the iSCSI server.

I have seen the same errors with iSCSI and ext4.

[2687538.144009] EXT4-fs (sdi): error count: 54
[2687538.144012] EXT4-fs (sdi): initial error at 1309736118: ext4_journal_start_sb:260
[2687538.144016] EXT4-fs (sdi): last error at 1309761117: ext4_put_super:737: inode 8194
[2774045.664009] EXT4-fs (sdi): error count: 54
[2774045.664013] EXT4-fs (sdi): initial error at 1309736118: ext4_journal_start_sb:260
[2774045.664016] EXT4-fs (sdi): last error at 1309761117: ext4_put_super:737: inode 8194
[2860553.184009] EXT4-fs (sdi): error count: 54
[2860553.184012] EXT4-fs (sdi): initial error at 1309736118: ext4_journal_start_sb:260
[2860553.184015] EXT4-fs (sdi): last error at 1309761117: ext4_put_super:737: inode 8194

I don't think that is a full jailbreak. It just lets you load your own applications. Plus if you want to distribute applications to regular phones you have to give them a 30% cut.

You don't have to start from scratch. You just have to enable the extents feature. It won't auto convert the old stuff, but any time something is changed it will be made into an extent.

Then your best bet would be to create a local repository out of the contents of cds, or a dvd. Which should be a basic thing you are going to do anyway if you have more than a few servers that don't have access to the internet. Then you would mirror in updates, and let them update from that.

There is graphical software that will let you install stuff straight from discs, and even ask for the right disc.

Is "yum install httpd" really that hard? I know I have done this before on plenty of servers.

I recently diagnosed two desktop machines. One ended up having a bad stick of memory, with the original symptoms being a corrupted copy of Windows XP that wouldn't boot. The other a bad hard drive, the symptoms being it would hang during use randomly and even during boot.

I used Prime95 and Memtest86+ to detect the bad stick of memory. Prime95 quickly came up with a error during the stress test, and Memtest86+ also came immediately came up with errors. In the past I have since subtle errors with Memtest86+ that only show up in later tests or with multiple passes. Instant answers isn't how it always goes.

For the bad hard drive I ended up doing a variety of tests. I tried Prime95, and since it was a Seagate drive, Seagate's Seatools. I didn't get any clear answers from them. At a later point I booted into a Fedora 11 Live cd, which popped up with a SMART error. Which ended up being a bad sector that needed to be remapped. I then tried using Spinrite to fix it, but ended up seeming to just hang on this one spot. So I replaced the hard drive. Afterword I reran Spinrite against the new drive, and it came up with nothing. I also played with Sandra Benchmarks at the end to stress the machine.

A lot of firmwares, like DD-WRT, have issues with binary only drivers and programs. I ran into it with the nas process in DD-WRT a few months ago.

  I had decided to move to WPA2 Enterprise. It sort of worked, but there is a long standing bug in DD-WRT relating to WPA2 Enterprise. WPA2 Enterprise depends on Radius. The nas process will only try a Radius server once. If it fails, then it won't try again. The only workaround is to kill the nas process one way or another. Then to make it all the more fun, the nas process is binary only.

  I ended up having to go back to the official Linksys firmware for my WRT600N to get working WPA2 Enterprise.

I wouldn't say it is snake oil. Putting versions in a page allows you to Google for it. Which makes the attack a lot easier. It also allows the attacker to do reconnaissance a lot less detectably a hold of time, and then spring it on everyone at once.

I have two setups like that.

  At home two 24" monitors on one computer, along with a second computer with a 20" monitor. They are connected with synergy and a ps/2 kvm. The kvm is good for when the main one is down. I can just hotkey over and use the second computer. I use it mainly for IM, but also sometimes for a second browser. Both computers are running Fedora I find having two computers comes in handy regularly. I also use the second computer as a iscsi server for the first. The first computer already has six drives in it. So the second computer allows me to expand to ten.

  The office setup is two 20" monitors on one computer, along with a second computer with a 20" monitor. They are connected with synergy. In this case I actually have two keyboards and nice. The main computer has no ps/2, and I have no usb kvms. I use a two port ps/2 kvm to share one keyboard between the second computer and a third computer. Then I toggle the monitor between dvi and vga. I do it this way since 99% of the time I don't need console access on the third computer. I access it via ssh for CUDA programs. The first computer runs Fedora, the second runs Vista, and the third runs Fedora. Vista is good in the office. It lets me do things like VMware Infrastructure client(currently Windows only), Internet Explorer(just today I was told to use IE on the HR site, since it works better), and other little things.

Sounds like it is either using MyISAM for tables, or you aren't using the --single-transaction option of mysql-dump for INNODB.